discussion / AI for Conservation  / 21 June 2022

Live Q&A on AI models to process Camera Trap Imagery: All about WildID

Hey everyone,

Do you have bulk camera trap imagery or are you looking to set up camera traps to collect wildlife data? Camera traps are undoubtedly one of the most useful and powerful tools used in wildlife monitoring, but you might want to consider how best you'll process the data collected. One way of sorting camera trap imagery, is to manually go through the 10,000 plus collected in order to group and classify the images, and the other, more efficient option is to use a technological tool; a trained AI/ML model to automatically sort, annotate and analyze the imagery. 

 Joining me in a few minutes , for a moderated live Q & A session in this thread, is Kate Tindall from the WildID team. Kate will take us through this trained AI/ML model that is used to process camera trap imagery, and will help us unpack the best ways to process data collected using camera traps, to have better output from our research studies/work. If you have any questions for Kate, please feel free to jump in and ask her during the Q&A session. 

A few things you need to know about the Live Q&A session: 

  1. The session will be moderated, but you can ask questions during and after the session e.g. If you need clarification on a question/answer, you are free to ask. 
  2. To ask a question that is linked to a specific answer, kindly press the ‘Reply to kate_alchemysoftware_co_za’ button, type in your question then save to post it.
  3. Before moving to each next question, we will give a ~3 minute window to allow you to ask any questions you have, but if you miss that window, you are totally okay asking the question afterwards.
  4. If you have any trouble viewing the questions or replies, please try and Refresh the page  

Welcome all to this discussion! I'm excited for what this session holds and hope you learn something new or get answers to some of the questions you have.
 

Netty.




Hello Kate. 

Thank you so much for finding time to join us for this session. Before we dive into the technical bit of discussing AI, camera traps and data, could you kindly introduce yourself in a few sentences?
 

Great introduction there! Tell us a bit about WildID. What is it and what are some of its long existing, new and upcoming features?


 

Porcupines, Kudus! Looks like there is a handful of species that can be detected using WildID! Before we find out about the specific species the model can so far detect, take us a few steps back and tell us the story behind this model? How did you come up with the idea of developing it?
 

I am sure many here can relate to the 1 million+ camera trap images or they know of a friends who have heaps of camera traps imagery lying around. 

So, I have a bulk of my camera trap imagery and would like to venture into training an AI model that will help me detect species and process my camera trap data. What does it involve to train my model, to achieve this?